Types vs. Sorts in HPSG

[Lutz Gunkel]

Hi Carl:

>I don't understand why one would want to think of feature DESCRIPTIONS
>as typed.

The reason, why I think that feature structure descriptions are sorted is
that I think this is just the correct way to interpret Pollard&Sag 1994 (as
well as other work in HPSG), especially with respect to the notion of
feature declaration. As I see it, a feature declaration fixes for any given
sort s which feature (names) f1, ..., fn are appropriate for s and which
sort the value of each feature (name) f1, ..., fn belongs to. Feature
declarations are thus part of the definition of the formal metalanguage in
defining the class of well-formed feature structure descriptions (or AVMs).
And as the well-formedness of a feature structure description depends on
the respective sort, I say that feature structure descriptions are sorted.
But maybe, then,  I am using the expression "sorted" in the wrong way
though I think that the issue is not a terminological one (and I am not
particularly interested in terminological issues either).

Best,
Lutz

____________________
Lutz Gunkel
FREIE UNIVERSITAET BERLIN
FB Philosophie und Geisteswissenschaften
Faechergruppe Germanistik
Habelschwerdter Allee 45
D-14195 Berlin
Tel.: +49(0)30 21751353  (home)
Email: gunkel at zedat.fu-berlin.de